SERLogic: A Logic-Integrated Framework for Enhancing Sequential Recommendations

Sequential recommendation models are used to predict users' next top-K preferred items based on their historical interactions. However, these models often struggle in "fuzzy areas" where recommendation scores are near decision thresholds, leading to false positives and false negatives...

Full description

Saved in:
Bibliographic Details
Published in:IEEE access Vol. 13; pp. 72221 - 72234
Main Author: Fan, Lihang
Format: Journal Article
Language:English
Published: Piscataway IEEE 2025
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Subjects:
Algorithms
Computational modeling
Logic
logic rules
Machine learning
Matrix decomposition
Motion pictures
parallel recommendation
Polynomials
Predictive models
Recommender systems
rule discovery
Sequential recommendation
Stars
Training
Transformers
Urban areas
ISSN:2169-3536, 2169-3536
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Sequential recommendation models are used to predict users' next top-K preferred items based on their historical interactions. However, these models often struggle in "fuzzy areas" where recommendation scores are near decision thresholds, leading to false positives and false negatives. To overcome this limitation, we propose <inline-formula> <tex-math notation="LaTeX">\textsf {SERLogic} </tex-math></inline-formula>, an innovative framework that incorporates logic rules, termed <inline-formula> <tex-math notation="LaTeX">\mathsf {TIE^{+}\!s} </tex-math></inline-formula>, into existing sequential recommendation models to enhance their accuracy without the need for training a new machine learning model. <inline-formula> <tex-math notation="LaTeX">\mathsf {TIE^{+}\!s} </tex-math></inline-formula> represent a novel class of graph prediction rules characterized by a dual graph pattern <inline-formula> <tex-math notation="LaTeX">\mathcal {Q} </tex-math></inline-formula> and a dependency <inline-formula> <tex-math notation="LaTeX">X \rightarrow (x, likes, y) </tex-math></inline-formula>, where <inline-formula> <tex-math notation="LaTeX">\mathcal {Q} </tex-math></inline-formula> exhibits a dual star structure, and X extends ML sequential recommendation models and 1-WL test as predicates. With <inline-formula> <tex-math notation="LaTeX">\textsf {SERLogic} </tex-math></inline-formula>, we show 1) validation problem for <inline-formula> <tex-math notation="LaTeX">\mathsf {TIE^{+}\!s} </tex-math></inline-formula> is in polynomial time (<inline-formula> <tex-math notation="LaTeX">\textsf {PTIME} </tex-math></inline-formula>), enabling efficient verification of whether a graph satisfies a set of <inline-formula> <tex-math notation="LaTeX">\mathsf {TIE^{+}\!s} </tex-math></inline-formula>; 2) creator-critic algorithm that iteratively learns high-quality <inline-formula> <tex-math notation="LaTeX">\mathsf {TIE^{+}\!s} </tex-math></inline-formula>; 3) parallel algorithm that applies the discovered <inline-formula> <tex-math notation="LaTeX">\mathsf {TIE^{+}\!s} </tex-math></inline-formula> to generate recommendations efficiently. Empirical evaluation on real-world datasets reveals that <inline-formula> <tex-math notation="LaTeX">\textsf {SERLogic} </tex-math></inline-formula> significantly enhances the performance of sequential recommendation models in terms of Recall@K and NDCG@K, while also achieving superior computational efficiency.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:2169-3536
2169-3536
DOI:10.1109/ACCESS.2025.3563977